pyckish 0.6.0

Pyckish is a micro framework for creating AWS Lambdas in Python in beautiful manner.

Pyckish

Create AWS Lambdas professionally with this micro framework

pip install pyckish

Pyckish is a micro framework for creating AWS Lambdas in Python in beautiful manner. It includes an "extract, parse and validate" solution for input, creation of exception handlers to deal with exceptions and better output, allowing you to return Pydantic models instead of dicts. One of Pyckish goals is to make using "Lambdas as handlers of HTTP requests" as an alternative that works similarly to other frameworks for back-end applications, like FastAPI.

Currently, it can be used to extract HTTP data that comes in the event/context dictionary. It extracts from the dictionary, parses it and validates it. It relies heavily on Pydantic, and will make your life simpler if you only like to deal with validated and correctly typed data.

Instead of doing this:

# No validation
# No parsing
# No exception handling
# No default value
def lambda_handler(event: dict, context: dict) -> dict:
    auth = event['headers']['authorization_token']
    store = event['pathParameters']['store']
    item = event['body']
    user = get_user(auth)
    similar_item: dict = get_similar_item(item, store, user)
    return similar_item

Do this:

import pyckish
from pyckish.http_elements import Body, Header, PathParameter
from my_models import Item


@pyckish.Lambda()
def lambda_handler(
        auth: str = Header(alias='authorization_token'),
        store: str = PathParameter(),
        item: Item = Body()
) -> float:
    user = get_user(auth)
    similar_item: Item = get_similar_item(item, store, user)
    return similar_item

And get validation and parsing free of trouble thanks to integration with Pydantic. Enjoy the advantages of a much more robust codebase, leaving behind having to extract and manage issues related to missing/wrong values.

What are AWS Lambda Functions

Lambdas are just simple functions that you can write in languages like Python, Javascript, etc. that are meant to be deployed on AWS. They can be activated/triggered by AWS whenever an event happens, it might a client application hitting on AWS API Gateway or a cron-job activation triggered by AWS Event Bridge. AWS manages every computer resource for you, this makes AWS Lambdas exceptionally easy to deploy. It is integrated with most things on AWS, meaning there is almost always a way to do what you want with Lambdas.

The interesting thing about Lambdas is that beyond being just a simple functions that are easy to write, easy to extract data from its inputs and put whenever logic you want inside it, Lambdas also are really, really cheap. You could activate it a million times per month without being charged, and this is only on AWS free tier. All of this makes Lambdas an attractive technology for most companies.

Lambdas, normally have only two parameters, both are two JSON's that are converted to python dictionaries, event and context. This library extracts, validates and parses values from those parameters.

Motivation

(Today, together with AWS API Gateway, it is possible to use only AWS Lambdas as back-end for your application. The problem is, unlike modern Frameworks, like FastAPI and Starlite, using only AWS Lambdas requires you to develop your own solutions for extracting, parsing, validating as well as creating error handling for the inputs of your code. There are solutions that allow you to use ASGI Frameworks with AWS Lambdas, like Mangum. But it is yet another technology that sits above your bulky framework. Personally, I think that the problem could be solved in a more simple and direct manner. Pyckish aims to be that solution.

Using tools like Serverless Framework with its integration with CloudFormation, many AWS Lambdas can be deployed from a single repository. Those "monorepos" solutions could also make heavy use of Pyckish in order to handle its inputs.

Right now, Pyckish is a tiny baby, and I'm not sure of its future. Weather it will become a full Framework with more capabilities than Chalice Framework, or it is going to remain as a simple "extractor/parser/validator" I do not know.

But I encourage you to try, simplicity and types will seduce you into it.

Usage

Currently, Pyckish provides you with these features:

• Extract/parse/validate the data contained in the Lambda Inputs all with adequate Exceptions raised in case of failure.
• Improves readability of your Lambda function by clearly stating the data your Lambda function requires.
• Allow you to add exception handlers to your Lambda in a easy manner, allowing to gracefully deal with exceptions.

Lambda Decorator

In order to pyckish to work it is required to add an instance of the Lambda class as a decorator above your lambda function. That is the only requirement. But in order to your function accept parameters you have to create some parameters in your function with type annotation.

Extract HTTP Data from Event

Pyckish provides classes that allows you to extract HTTP Data from the event, such ass Path, Method, PathParameter, PathParameters, Header, Headers, QueryParameter, QueryParameters and Body. These classes are all children classes of LambdaInputElement class.

The version in the singular means they are going to extract only one parameter. They require a type annotation that it the type of that specific parameter. The ones in the plural means they are going to extract all parameters at once, the type annotation needs to be a Pydantic Model.

Checkout the difference:

import pyckish
from pyckish.http_elements import Header, Headers
import pydantic


class MyAuthHeader(pydantic.BaseModel):
    auth: str = pydantic.Field(alias='authorization')
    host: str


event = {
    'headers': {'authorization': 'token', 'host': '177.177'}
}
context = {}


@pyckish.Lambda()
def lambda_handler(
        auth: str = Header(),
        my_header: MyAuthHeader = Headers()
) -> None:
    print(auth)
    print(my_header)


lambda_handler(event, context)

Simple extraction from Event

If you do not provide a child of LambdaInputElement class on the default value, the name of the parameter act as a key to be extracted on the event.

import pyckish

event, context = {'my_value_on_event': '200'}, {}


@pyckish.Lambda()
def lambda_handler(
        my_param_on_event: int = 500
) -> None:
    print(f'value: {my_param_on_event}, type: {type(my_param_on_event)}')


lambda_handler(event, context)

Custom LambdaInputElement

If you want to extract your own value from the event or context with validation and parsing capabilities, you can create your own LambdaInputElement child class. It is required that this class implements a method called "extract" accepting a LambdaInput instance. LambdaInput is just a dataclass with two attributes, event and context.

import pyckish
from pyckish import LambdaInputElement
from pyckish import ValidationError


class MySpecialParameter(LambdaInputElement):
    def extract(self, event: dict, context: dict) -> str:
        try:
            return event['my_special_parameter_key']['another']
        except KeyError:
            raise ValidationError('My special parameter is missing')


@pyckish.Lambda()
def lambda_handler(
        param: str = MySpecialParameter()
) -> None:
    print(f'my param: {param}')

Adding Exception Handlers

Exception Handlers are functions to be executed when an error occurs in your lambda. To make a function to each error use the add_exception_handler method.

import pyckish


class MyException(Exception):
    pass


# This signature is required
def handler_for_my_exception(event: dict, context: dict, exception: Exception) -> str:
    # this return is going to be the lambda's return value
    return 'My exception occurred'

@pyckish.Lambda(
  exception_to_handler_mapping={MyException: handler_for_my_exception}
)
def lambda_handler() -> None:
    raise MyException()


lambda_handler({}, {})

Formatting Response to be adequate with AWS API Gateway

Pyckish is also capable of formatting your lambda response to what AWS API Gateway expects, ir order to send it as HTTP data. AWS API Gateway expects a Body, Headers and StatusCode field in the event dictionary in order to send it as a response to its client. By using the flag, is_http=True in the Lambda decorator. Anything passed as a response will be sent as Body parameter to AWS API Gateway, you can also specify the status code directly on the decorator (this status code will be used in case of success) or by specifying a HTTPResponse object as a return of your function. Headers can also be specified in the HTTPResponse object.

Add inbound and outbound interceptors

...